Failure Mechanism and Force-Displacement Relationship of Backfill behind Rigid Retaining Wall under Passive Condition

Abstract

Earth-retaining structures are the most important and inevitable structural components of almost every civil-infrastructural project, such as bridge abutments, approach embankments, grade separation, anchor blocks, pile caps, culvert walls, tunnel portals, bulkheads, and retaining walls. Investigations into the progressive development of mobilized soil mass, the onset of strain localization, and the development of ultimate failure in the backfill behind the earth retaining structures are not only critical but are necessary. This is because the information gathered from such investigations is essential to validate the assumptions made in the prevailing force-based analytical models about the geometric characteristics of the ultimate failure of the surface and to pave the way for the application of the most advanced displacement-based design concepts. In the current study, an effort has been made to explore the progressive development of the backfill zone when the wall was moved monotonically towards the soil in static state conditions.

In the present study, the reduced-scale laboratory model of the retaining wall test integrated with the image-capturing assembly was developed to visualize the incremental features of the mobilized soil mass and passive sliding surfaces with the assistance of the digital image correlation technique. This test matrix has been divided into two groups. In group 1, two backfill materials (i.e., sand and fine gravels) were compacted at seven different backfill surface inclinations (β= -15⁰, -10⁰, -5⁰, 0⁰, +5⁰, +10⁰, +15⁰) in medium dense state conditions. In group 2 of the test matrix, a series of experimental tests have been performed, disclosing the progressive failure mechanism in the backfill material compacted at different density levels. The poorly graded sand and fine gravels are compacted in loose, medium-dense, and dense state conditions. The earth pressure coefficients obtained in the retaining wall tests were compared with the values obtained from the analytical models. The successive images of the deformed backfill were analyzed through Vic-2D software for the development of displacement and strain contours.

Before conducting the Digital Image Correlation (DIC) analysis of the recorded images, the results of DIC analysis were validated by generating the finite element model of the wall setup. The image analysis successfully captured the evolution of mobilized soil mass and the development of passive failure surfaces. The image analysis of group 1 specimens disclosed the presence of a distinct boundary within the backfill, beyond which soil particles remained immobile during the pushover test. This boundary manifested during the early phases of the test when stress levels were relatively low. For specimens with positive slopes, this boundary evolved into the ultimate failure surface, characterized by a geometry resembling a log-spiral curve. Conversely, for specimens with negative slopes, the failure surface may not adhere to this log-spiral boundary; instead, it might follow a more direct route, resembling the straight line predicted by Rankine theory. The DIC data of group 2 specimens reveals that the phenomenon of strain localization was delayed in the loose state condition of the specimen, and the yielding length of the strain band is much lower in the loose state condition as compared to the specimen in the dense state condition.

A coherent approach to modify the Log-Spiral Hyperbolic (LSH) model was proposed in the current study. This approach was used to predict the nonlinear passive capacity-displacement relationship of the backfill which will be used as essential input data for advanced analysis methods and paved the way for implementing advanced next-generation displacement-based models. This study presents a new analytical method for converting the stress Mohr circle into the strain Mohr circle, because of the lack of a trustworthy nonlinear plane-strain constitutive model for soil in the literature to date. This technique was created especially to deal with plane-strain geotechnical problems. To estimate the effective Young's and shear moduli, the aforementioned approach takes into consideration the nonlinear character of soil without deviating from the core principles of Poisson's effect. It performs this by incorporating empirical stiffness reduction relationships. After the strain distribution across the backfill has been accurately identified, a more sophisticated strain integration system that takes into account the contributions from both shear and normal stresses is created for displacement computations. Furthermore, we suggest renaming the term "intermediate failure surface" to "shear strain concentration belt". This is because, in the pre-failure state, the shear stress levels are still too low to establish a complete sliding plane within the backfill. Furthermore, the shear strain concentration belt's beginning point is always at the base of the wall. Before implementing Mohr’s circle conversion method to predict the force-displacement curves, the method was validated by comparing the strain data obtained from the proposed approach with the strain values obtained for the FE model of the retaining wall test.

The proposed model required inputting 8 parameters, two of which are wall dimensions including height (H) and width (B), four of which are backfill soil properties including soil unit weight (γ_s), relative density (D_r), wall-soil friction angle (δ_w), and shear strength parameters (φ and c), two of which are seismic coefficients (k_h, k_v) and 3D correction factor (R). The proposed approach was validated by comparing the results with the experimentally reported values in terms of force-displacement curves.

Date of Award	12 Sept 2024
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Yiu Yin Raymond LEE (Supervisor) & Shi-Yu XU (Supervisor)